Implementations directed to providing a computer-implemented method comprising receiving, from a motion sensing device, a plurality of image frames that includes information regarding a plurality of animals housed in a group-housed environment, determining a coordinate space of the group-housed environment based on an analysis of a first image frame of the image frames, generating, based on the analysis of the first image frame, an ellipsoid model for each animal based on defined surface points for each animal weighted according to a likely proximity to a crest of a spine of the respective animal, and tracking a position and an orientation of each animal within the image frames by enforcing shape consistency of the ellipsoid models, and adjusting the position of each of the ellipsoid models based on the defined surface points for each animal and a maximum likelihood formulation of a movement distance for each animal.
Legal claims defining the scope of protection, as filed with the USPTO.
1. A computer-implemented method comprising: receiving, from a motion sensing device, a plurality of image frames that includes information regarding a plurality of animals housed in a group-housed environment; determining a coordinate space of the group-housed environment based on an analysis of a first image frame of the image frames; generating, based on the analysis of the first image frame, an ellipsoid model for each animal based on defined surface points for each animal weighted according to a likely proximity to a crest of a spine of the respective animal, in which a first surface point that is more likely to be near the crest of the spine of the respective animal is given a higher weight than a second surface point that is less likely to be near the crest of the spine of the respective animal; and tracking a position and an orientation of each animal within the image frames by: enforcing shape consistency of the ellipsoid models; and adjusting the position of each of the ellipsoid models based on the defined surface points for each animal and a maximum likelihood formulation for each animal.
2. The method of claim 1 , wherein the maximum likelihood formulation alternates between assigning maximum-likelihood clusters to the defined surface points for each animal via a metric that enforces an ellipsoidal distribution, and a recalculation of parameters of the maximum-likelihood clusters.
3. The method of claim 1 , wherein tracking the position and the orientation of each animal includes a use of an adaptive formulation of an exponential smoothing tracker.
4. The method of claim 1 , wherein determining the coordinate space of the group-housed environment is based on annotating corner points of the group-housed environment in the first image frame.
5. The method of claim 4 , wherein annotating the corner points of the group-housed environment is performed via a user interface.
6. The method of claim 4 , wherein the first image frame is further annotated to isolate points in a foreground of the first image from a background of the first image by removing points that lie outside bounds of the group-housed environment bounds and masking out points that lie in an image space of a feeder and a waterer.
7. The method of claim 1 , wherein the position and the orientation of each animal is characterized by a centroid and rotation matrix applied to each respective ellipsoid model.
8. The method of claim 1 , wherein adjusting the position of each of the ellipsoid models within each image frame is based on recalculating a centroid and an orientation of each respective ellipsoid model by weighting the respective surface points according to a likely proximity to the crest of the spine of the respective animal.
9. The method of claim 1 , wherein the animals are pigs.
10. The method of claim 1 , wherein the motion sensing device is a depth camera.
11. The method of claim 1 , comprising providing the tracking the position and the orientation data to a livestock management system that tracks health, wellbeing, growth-rate, and aggression among the animals in the group-housed environment.
12. The method of claim 1 in which adjusting the position of each of the ellipsoid models comprises adjusting the position of each of the ellipsoid models based on the defined surface points for each animal and a maximum likelihood formulation of a movement distance for each animal.
13. One or more non-transitory computer-readable storage media coupled to one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for tracking a plurality of animals, the operations comprising: receiving, from a motion sensing device, a plurality of image frames that includes information regarding the animals housed in a group-housed environment; determining a coordinate space of the group-housed environment based on an analysis of a first image frame of the image frames; generating, based on the analysis of the first image frame, an ellipsoid model for each animal based on defined surface points for each animal weighted according to a likely proximity to a crest of a spine of the respective animal, in which a first surface point that is more likely to be near the crest of the spine of the respective animal is given a higher weight than a second surface point that is less likely to be near the crest of the spine of the respective animal; and tracking a position and an orientation of each animal within the image frames by: enforcing shape consistency of the ellipsoid models; and adjusting the position of each of the ellipsoid models based on the defined surface points for each animal and a maximum likelihood formulation for each animal.
14. The non-transitory computer-readable storage media of claim 13 , wherein the maximum likelihood formulation alternates between assigning maximum-likelihood clusters to the defined surface points for each animal via a metric that enforces an ellipsoidal distribution, and a recalculation of parameters of the maximum-likelihood clusters.
15. The non-transitory computer-readable storage media of claim 13 , wherein tracking the position and the orientation of each animal includes a use of an adaptive formulation of an exponential smoothing tracker.
16. The non-transitory computer-readable storage media of claim 13 , wherein determining the coordinate space of the group-housed environment is based on annotating corner points of the group-housed environment in the first image frame.
17. The non-transitory computer-readable storage media of claim 16 , wherein the first image frame is further annotated to isolate points in a foreground of the first image from a background of the first image by removing points that lie outside bounds of the group-housed environment bounds and masking out points that lie in an image space of a feeder and a waterer.
18. The non-transitory computer-readable storage media of claim 13 , wherein the position and the orientation of each animal is characterized by a centroid and rotation matrix applied to each respective ellipsoid model.
19. The non-transitory computer-readable storage media of claim 13 , wherein adjusting the position of each of the ellipsoid models within each image frame is based on recalculating a centroid and an orientation of each respective ellipsoid model by weighting the respective surface points according to a likely proximity to the crest of the spine of the respective animal.
20. The non-transitory computer-readable storage media of claim 13 , in which adjusting the position of each of the ellipsoid models comprises adjusting the position of each of the ellipsoid models based on the defined surface points for each animal and a maximum likelihood formulation of a movement distance for each animal.
21. A system, comprising: a motion sensing device; one or more processors; and a computer-readable storage device coupled to the one or more processors and having instructions stored thereon which, when executed by the one or more processors, cause the one or more processors to perform operations for tracking a plurality of animals in a group-housed environment, the operations comprising: receiving, from the motion sensing device, a plurality of image frames that includes information regarding the animals housed in the group-housed environment; determining a coordinate space of the group-housed environment based on an analysis of a first image frame of the image frames; generating, based on the analysis of the first image frame, an ellipsoid model for each animal based on defined surface points for each animal weighted according to a likely proximity to a crest of a spine of the respective animal, in which a first surface point that is more likely to be near the crest of the spine of the respective animal is given a higher weight than a second surface point that is less likely to be near the crest of the spine of the respective animal; and tracking a position and an orientation of each animal within the image frames by: enforcing shape consistency of the ellipsoid models; and adjusting the position of each of the ellipsoid models based on the defined surface points for each animal, a maximum likelihood formulation for each animal.
22. The system of claim 21 , wherein the position and the orientation of each animal is characterized by a centroid and rotation matrix applied to each respective ellipsoid model.
23. The system of claim 21 in which adjusting the position of each of the ellipsoid models comprises adjusting the position of each of the ellipsoid models based on the defined surface points for each animal and a maximum likelihood formulation of a movement distance for each animal.
24. The system of claim 21 in which the adjusting the position of each of the ellipsoid models comprises a recalculation of a centroid and an orientation of each respective ellipsoid model by weighting the respective surface points according to a likely proximity to the crest of the spine of the respective animal.
Cooperative Patent Classification codes for this invention. Click any code to explore related patents in that topic.
August 28, 2018
October 6, 2020
Browse 5M+ US patents with plain-English claim translations and AI-generated analysis.